"EL" calculation: if input contains NaN?

[DJ4seasons]

For the example in https://unidata.github.io/MetPy/v1.4/api/generated/metpy.calc.el.html#metpy.calc.el,
if one elemnet of rh is changed to zero (it happens at least with MERRA2), calculated Td contains a "NaN."
With Td containing a "NaN", el stops working and shows error:

Traceback (most recent call last):
File "", line 1, in
File "/Users/djin1/opt/miniconda3/lib/python3.8/site-packages/metpy/xarray.py", line 1329, in wrapper
result = func(*bound_args.args, **bound_args.kwargs)
File "/Users/djin1/opt/miniconda3/lib/python3.8/site-packages/metpy/units.py", line 320, in wrapper
return func(*args, **kwargs)
File "/Users/djin1/opt/miniconda3/lib/python3.8/site-packages/metpy/calc/thermo.py", line 889, in el
x, y = find_intersections(pressure[1:], parcel_temperature_profile[1:], temperature[1:],
File "/Users/djin1/opt/miniconda3/lib/python3.8/site-packages/metpy/xarray.py", line 1329, in wrapper
result = func(*bound_args.args, **bound_args.kwargs)
File "/Users/djin1/opt/miniconda3/lib/python3.8/site-packages/pint/registry_helpers.py", line 283, in wrapper
result = func(*new_values, **kw)
File "/Users/djin1/opt/miniconda3/lib/python3.8/site-packages/metpy/calc/tools.py", line 136, in find_intersections
nearest_idx = nearest_intersection_idx(a, b)
File "/Users/djin1/opt/miniconda3/lib/python3.8/site-packages/metpy/calc/tools.py", line 87, in nearest_intersection_idx
difference = a - b
ValueError: operands could not be broadcast together with shapes (29,) (28,)

This error looks like because nan was masked in Td, but other inputs still have the same dimension.
I suggest that it would be great if the calculation modules include a function that

1. check if any input data contains nan value
2. if nan exists, masking all the input arrays at the location of nan, so all input data are always of same length.

[dopplershift]

That's a great idea...which is why the current implementation of el has the following line:

MetPy/src/metpy/calc/thermo.py

Line 875 in c940bc0

pressure, temperature, dewpoint = _remove_nans(pressure, temperature, dewpoint)

😉

Can you share the full (preferrably minimal) arrays of pressure, temperature, and dewpoint that reproduce your problem so that we can see what's actually breaking down and causing the failure?

[DJ4seasons]

I produced the error by using values in example (https://unidata.github.io/MetPy/v1.4/api/generated/metpy.calc.el.html#metpy.calc.el)
What I did was simply,
(after copy and paste p, T, and rh)

rh[-2]=0

then I followed next steps, calculate Td, prof, and el.

Thanks
Daeho Jin

---

Environment:
Python 3.8.15 | packaged by conda-forge | (default, Nov 22 2022, 08:49:06)
[Clang 14.0.6 ] on darwin
Type "help", "copyright", "credits" or "license" for more information.

import metpy
metpy.version
'1.4.0'

[DJ4seasons]

It looks that the "_remove_nans()" remove nans for p, T, and Td.
However, when I provide "parcel_profile" like the example, it is not touched, so I think the error occurs because of the difference of length between prof and other variables.

[dopplershift]

Good catch. That's exactly the problem (and for lfc as well). I've opened #2921 to track this and remind us to fix it for the next release. If you're feeling particularly motivated, feel free to submit a PR with the fix and some tests.